Background technology
Sun sensor is the important devices of carrying out attitude measurement on the spacecraft, has been widely used in space industry now, comprises space shuttle, earth satellite and deep space probe etc.In recent years,, spacecraft with respect to the attitude measurement accuracy of the sun with to measure the requirement of reliability more and more higher, is especially needed the system of high precision imaging, comprise investigation satellite, resource and astronomical sight satellite and telstar along with the requirement of space mission.High-precision angle of incidence of sunlight degree measure can realize the spacecraft solar array accurately to day, obtain with the ceiling capacity that guarantees satellite.Simultaneously, based on the angle of incidence of sunlight degree measure to day stable mode also be a kind of originate mode and the safe mode of most satellites.
The photosensitive detecting element of traditional analog sun sensor adopts the solar battery sheet based on photovoltaic property, analog quantity output, and ratio of precision is lower, and anti-interference is not strong yet, can not satisfy the big visual field and the high-precision requirement of spacecraft attitude control.Along with the development of CCD technology, also occur based on the digital sun sensor of line array CCD, and on aerospace engineering, use gradually, its precision improves with respect to analog sun sensor.
As shown in Figure 1, in digital sun sensor was analyzed, imaging model that it is generally acknowledged sun sensor was the pinhole imaging system model, i.e. t of a certain moment, and sunshine obtains unique imaging point coordinate (x through light introducer imaging on detector of sun sensor
t, y
t).The focal length of supposing sun sensor simultaneously is f, obtains the inertia vector v of the sun like this
tVector is in the sun sensor coordinate system
(x wherein
0, y
0) be the central point of sun sensor in the sunshine zero angle.
Under normal conditions, we are with t x axle angle of incidence of sunlight α constantly
tWith y axle angle of incidence of sunlight β
tBe expressed as respectively:
F, x in following formula
0, y
0Be system constants, so diaxon angle of incidence of sunlight (α
t, β
t) precision depend mainly on (x
t, y
t) computational accuracy.Existing digital sun sensor generally adopts the light introducer of single hole or single seam formula, only can obtain a picture point (x on sensitive detector
t, y
t), the precision of sun sensor only depends on the computational accuracy of single hole.Stopped up by the space material when aperture is interfered, perhaps any variation takes place in the pattern of aperture, all can cause the precision of sun sensor to descend fast, even causes sun sensor to lose efficacy.And when imageing sensor because environmental problem such as space irradiation when causing small number of pixels to be damaged, if the imaging point of monotrysian type sun sensor is when appearing at failure area, also can cause it can not operate as normal.
Summary of the invention
At the problems referred to above, the purpose of this invention is to provide a kind of high-precision APS solar sensor that solves the whole sun sensor cisco unity malfunction problem that causes by the computational accuracy misalignment of single hole.
For achieving the above object, the present invention takes following technical scheme: a kind of high-precision APS solar sensor is characterized in that: it comprises light introducer, imageing sensor, interface circuit and computing machine; The window of described smooth introducer is arranged in order by substrate of anti-irradiation quartz glass and sunray barrier bed to be formed, and described sunray barrier bed comprises the light hole that is arranged in p * p matrix; Described imageing sensor has big face battle array, and can carry out window explosion and read at random; Described interface circuit is imported the image information of described imageing sensor output in the described computing machine and is calculated.
Described imageing sensor is the APS cmos image sensor.
A kind of implementation method of high-precision APS solar sensor, it may further comprise the steps:
1) obtains the standard form image
Simulate real running environment in the laboratory, make simulated solar irradiation vertically inject the light introducer of sun sensor, the image that this moment, imageing sensor collected is as the standard form image; Described standard form image comprises a p * p imaging point matrix; According to the gray-scale value of the pixel of the imaging point of described standard form image, calculate the centroid position (x of the imaging point matrix of described standard form image
0, y
0);
2) read the part of imaging point matrix image in the true environment
When a) for the first time starting sun sensor, the image that described imageing sensor is collected travels through searches imaging point matrix center (x when the record initial pixel reads
1, y
1); B) after described initial pixel reads, according to definite imaging point matrix center (x of last time
1, y
1) determine the imaging point estimation range when current time t sun sensor upgrades, write down the center (x of each imaging point estimation range
k, y
k), k=1 ..., p * p; C) according to the center (x of described each imaging point estimation range
k, y
k), local imaging is carried out and image is read in each imaging point estimation range, extract the imaging point estimation range image of described current time t;
3) imaging point matrix image related operation and the computing of correlated results barycenter
I) to the gray-scale value of the pixel of each imaging point of the gray-scale value of the pixel of each imaging point of described imaging point estimation range image and described standard form image, carry out related operation, obtain correlation matrix; Ii) described correlation matrix is carried out the barycenter computing, obtain imaging point barycenter (x '
k, y '
k), and then obtain the relative barycenter (x of imaging point matrix by weighted mean filtering
c, y
c); Iii) according to described imaging point matrix center (x
1, y
1) with the relative barycenter (x of described imaging point matrix
c, y
c), obtain the barycenter exact position of current time t imaging point matrix:
x
t=x
1+x
c
y
t=y
1+y
c
4) calculate angle of incidence of sunlight
Barycenter exact position (x according to described imaging point matrix
t, y
t) and the centroid position (x of described imaging point matrix
0, y
0), calculate the incident angle α of current time t sunshine with respect to described sun sensor
t, β
t:
Wherein, f is the renewal frequency of described sun sensor.
In the described step 1), the gray-scale value according to the pixel of the imaging point of described standard form image by the weighted mean filtering algorithm, calculates the centroid position (x of the imaging point matrix of described standard form image
0, y
0).
In the described step a), described image traveled through every the row pixel according to interlacing search.
Described step I i) in, when the situation that exists imaging point to lose, the weight coefficient of the described weighted mean filtering of the imaging point correspondence of then losing is 0.
The present invention is owing to take above technical scheme, it has the following advantages: 1, the light introducer of sun sensor of the present invention is owing to comprise a plurality of light holes that are arranged in matrix, therefore can produce than the more imaging point of single hole sun sensor, sun sensor of the present invention is to each imaging point process image related operation and the computing of correlated results barycenter simultaneously, and the correlation matrix barycenter is weighted mean filter, degree of accuracy improves several times than single imaging point method.2, because the present invention adopts the APS cmos image sensor as sensitive detector, the ability that has window explosion and read at random, not only can avoid the readout time and the long problem of processing time of big array image sensor, also effectively utilize the resource of imageing sensor, improved the work efficiency of image acquisition and treatment circuit.3, because smooth introducer of the present invention has a plurality of light holes, and have big spacing between the aperture, therefore can effectively avoid noise effects such as imageing sensor top pixel damage, improve the reliability and stability of sun sensor.High-precision APS solar sensor of the present invention can be widely used in attitude measurement control system on the spacecraft.
Embodiment
Below in conjunction with drawings and Examples the utility model is described in detail.
As shown in Figure 2, active pixel sensor of the present invention (APS, Actel Pixels Sensor) sun sensor is a porous array type, and it comprises light introducer 1, imageing sensor 2, interface circuit (not shown) and computing machine (not shown).Wherein the window of light introducer 1 is arranged in order by substrate of anti-irradiation quartz glass and sunray barrier bed and forms, and the sunray barrier bed comprises p * p=k the light hole that is arranged in matrix, is used to produce the spot array image.Imageing sensor 2 is the APS cmos image sensor 2 that has certain distance with light introducer 1, it has big face battle array, and can carry out window explosion and read at random, be used for each grey scale pixel value of output facula array image, and each pixel has the dynamic range of 60dB at least.Interface circuit is used for the computing machine of the image information input sun sensor of imageing sensor 2 outputs is calculated.When sunray incided sun sensor with different angles, the image that light produces on imageing sensor 2 by light introducer 1 comprised a p * p imaging point matrix, and each imaging point comprises q * q pixel.
The imaging characteristics of sun sensor according to the present invention, each imaging point approximately occupies 5 * 5 pixels, so work as p=6 in the present embodiment, during q=5, k=36 imaging point occupies about 900 pixels altogether. with respect to entire image sensor 2 totally 100 ten thousand pixels, all imaging points only account for about per mille. because the imageing sensor 2 of big battle array of the present invention is when improving single pixel resolution, also increased the workload of overall calculation, therefore the present invention can improve the turnover rate of system greatly by the method that adopts the pixel region partial exposure and read, thereby improve the processing speed of system, also provide assurance for estimation range of the present invention and image related algorithm.
APS sun sensor of the present invention realizes that the step of spacecraft attitude measurement is as follows:
1, obtains the standard form image
This step real running environment to sun sensor in the laboratory is simulated, and major equipment comprises solar simulator, electronic theodolite, two-axle rotating table and sun sensor.At first, use electronic theodolite that solar simulator is measured, guarantee solar simulator and electronic theodolite point-blank.Secondly, sun sensor is installed on the two-axle rotating table, rotate two-axle rotating table, make sun sensor, solar simulator and electronic theodolite point-blank, make simulated solar irradiation vertically inject light introducer 1, guaranteed incident angle α=0 of simulated solar irradiation, β=0, wherein, α, β are expressed as the position angle and the ascending angle of sun sensor respectively.As shown in Figure 3, α=0, the image that β=0 o'clock imageing sensor 2 collects is as standard form image T.
Since embodiments of the invention be 6 * 6 porous array type sun sensors it have the light introducer 1 that comprises 6 * 6 light holes, therefore can from standard form image T, extract 6 * 6 imaging point matrixes of forming by 36 imaging points.As shown in Figure 4, utilize the above-mentioned standard form image T that obtains to extract effective image-region and valid pixel, consider according to 5 * 5 pixels for each imaging point, when sunshine projects on the imageing sensor 2 by light introducer 1, the gray-scale value T of each effective imaging point pixel among the imageing sensor 2 outputting standard template image T
k(i, j), k=1 wherein ..., 36, i, j=1 ..., 5.Afterwards according to gray-scale value T
k(i, j) centroid position (x of each imaging point of basis of calculation template image
0k, y
0k), by the weighted mean filtering algorithm, obtain the centroid position (x of standard form image imaging dot matrix
0, y
0).The image information of standard form image T is preserved in the computing machine of sun sensor, in order under true environment, reading.
2, read the part of imaging point matrix image in the true environment
1) initial pixel reads
When for the first time starting sun sensor under true environment, the particular location that imaging point drops on the imageing sensor 2 of sun sensor is unknown, therefore need travel through the entire image that imageing sensor 2 is gathered and search.In order to accelerate the seek rate of imaging point, entire image is searched every the row pixel according to interlacing.Because the distance between each imaging point is fixed, therefore after finding first imaging point, can utilize the relation between each imaging point, directly search and judge whether all the other 35 imaging points are correct.The aperture of considering light introducer 1 exists by the situation that the space material stops up or pattern changes, and when finding imaging point correct more than 30, can represent that initial pixel reads process and completes successfully.Imaging point matrix center (x when the record initial pixel reads
1, y
1).
2) be specified to the picture point estimation range
Because the renewal frequency f of sun sensor is than higher, about 30Hz, the difference of the imaging point position on imageing sensor 2 that obtains after twice continuous renewal is very little.Therefore, after initial pixel reads the later renewal of sun sensor each time, according to definite imaging point matrix center (x of last time
1, y
1) determine the imaging point estimation range that current time upgrades.After sun sensor upgrades, serve as that the basis is extended to four limits and searched with the imaging point position of last time, 4 pixels are extended on every limit, obtain each imaging point estimation range, and each imaging point estimation range comprises 13 * 13 pixels.After record current time t sun sensor upgrades, the center (x of imaging point estimation range
k, y
k), k=1 wherein ..., 36 imaging points of 36 expressions.
3) be extracted into picture point estimation range image
After current time t sun sensor upgrades, the center (x of determined estimation range, imaging point position when upgrading according to the last time
k, y
k), the characteristic of reading in conjunction with the random exposure and the window of APS cmos image sensor 2, to carrying out local imaging and image is read in each imaging point estimation range, extract current time t imaging point estimation range image P, each imaging point of imaging point estimation range image P comprises 13 * 13 pixels.P
k(u v) represents the gray-scale value of the pixel of each imaging point among the image P of estimation range, k=1 wherein ..., 36 imaging points of 36 expressions, u, v=1 ..., 13 are expressed as each locations of pixels coordinate in the picture point estimation range.
3, imaging point matrix image related operation and the computing of correlated results barycenter
I) to the gray-scale value P of the pixel of each imaging point among the image P of above-mentioned imaging point estimation range
k(u, v) with standard form image T in the gray-scale value T of pixel of each imaging point
k(i j) carries out the total correlation computing respectively, obtains correlation matrix C
k, C
k9 * 9 correlation matrixes of representing k imaging point.As shown in Figure 5, from imaging point estimation range image 3, the relevant reference position 4 (m=n=0) of imaging point calculates the relevant end position 5 (m=n=8) of imaging point always, and accounting equation is as follows:
Correlation matrix C
kOperation result as shown in Figure 7.
Ii) to correlation matrix C
kCarry out the barycenter computing, obtain 36 imaging points barycenter (x '
k, y '
k), again with each (x '
k, y '
k) be weighted mean filter, obtain the relative barycenter (x of imaging point matrix
c, y
c), concrete is calculated as follows:
(4)
(5)
Wherein, a
k, b
kThe expression weight coefficient has following relation:
Under normal conditions, it is impartial that we get the weight coefficient of 36 imaging points, i.e. a
k=b
k=1/36.When the situation that exists imaging point to lose, for the sun imaging point that guarantees to lose to not influence of overall calculation, the weight coefficient of this moment is:
Equation (6) shows that if certain light hole stops up, the weight coefficient of its pairing imaging point is 0, i.e. a
k=b
k=0, guarantee that like this it can not exert an influence to overall precision, the problem that has solved little hole plug and lost.
Imaging point matrix center (x when iii) reading according to initial pixel
1, y
1) relative barycenter (x with the imaging point matrix
c, y
c), obtain the exact position of current time t imaging point matrix:
x
t=x
1+x
c
(7)
y
t=y
1+y
c
4, calculate angle of incidence of sunlight
According to angle of incidence of sunlight α, the barycenter exact position (x of β and imaging point matrix
t, y
t) and the centroid position (x of standard form image imaging dot matrix
0, y
0) relation, in conjunction with equation in the background technology (1), equation (2), calculate the incident angle α of current time t sunshine with respect to sun sensor
t, β
t:
F, x in equation (7) and equation (8)
0And y
0Be system constants, so angle of incidence of sunlight (α
t, β
t) precision depend primarily on imaging point centroid position (x
t, y
t) computational accuracy.
The present invention has adopted the method for 36 sun imaging point weighted means filtering to come accuracy of measurement system, supposes that the error of calculation noise equivalent angle of single imaging point is σ
0, and angle of incidence of sunlight α, the precision of β directly is decided by σ
0When adopting 36 imaging points, suppose that equally the error of calculation noise equivalent angle of each picture point is σ
0, be about through filtered error of calculation noise equivalent angle:
Therefore can think that the angle of incidence of sunlight degree of accuracy of sun sensor improves 6 times than single imaging point method, and for the APS CMOS sensitive detector that adopts 1024 * 1024, its precision can be up to 0.01 degree level.